That Field-Strengh Meter again...

[Fred Bartoli]

"Paul Burridge" <pb@osiris1.notthisbit.co.uk> a écrit dans le message news:
1h3e005f2drsbs05h498pkj9g2r0q88u5v@4ax.com...

On Thu, 15 Jan 2004 19:21:07 +0100, "Fred Bartoli"
fred._canxxxel_this_bartoli@RemoveThatAlso_free.fr_AndThisToo> wrote:


"Frank Bemelman" <fbemelx@euronet.invalid.nl> a écrit dans le message
news:
4006cf94$0$125$3b62cedd@news.wanadoo.nl...
"Paul Burridge" <pb@osiris1.notthisbit.co.uk> schreef in bericht
news:6i9d00ts1mup1br954si3llamb4qhk2cp1@4ax.com...
On Thu, 15 Jan 2004 12:28:56 +0100, "Frank Bemelman"
fbemelx@euronet.invalid.nl> wrote:

Okay, but I still don't know what it is.


Guess what ?
Neither Paul, nor every body here.

There's simply nothing like that.

No, no, no. I'm gonna get me a Nobel Prize for this one. I just don't
want it (this thread's deviating) upsetting the current question and
thereby robbing me of a solution to the FSM problem.
--

Unfortunately that's the kind of pb that may be easily diagnosed and solved
by the knowledgeable peoples when having the beast in hands but *much*
harder at distance.

Check your proto against elementary and good wiring practice :
Ground plane, good supply decoupling/bypass, short tracks and leads
(specially for bypass caps)
low inter stage to stage parasitics coupling (capacitive, inductive and via
power rail). If you've used a ground plane, gnd ref should not be a pb for
such a simple design unless you've done a VERY VERY BIG MISTAKE.

Hmm, wait a minute. Looking at the schematics and, ahem, knowing you, how
did you handled the supply rail ?
Did you only provided a pair of 10n caps right at the batteries as shown on
the schematics ?
If the answer is yes, never forget that real wires are never near from the
perfect ones in a sim.
Make sure you've carefully bypassed the nodal point between L1 and L2.
Put a 22n to 100n cap right at top of L1 and L2 with possibly a 10R resistor
between each and keep lead short.

from meter
stage supply
|
|
+---||-+----10R-----+----||-+
GND ) ) GND
)L1 )L2
) )



Fred.

 

[Walter Harley]

"Paul Burridge" <pb@osiris1.notthisbit.co.uk> wrote in message
newsq8d00d2fbmgv96kb8nmpmct473griocd6@4ax.com...

How are you keeping L1 from talking to L2?

I've tried interposing a grounded copper plate between them but it
made no difference. Pity as it was a good suggestion.

No, it was not a good suggestion.

A grounded copper plate would shield against electrical field, not magnetic
field. Proximate inductors couple via magnetic fields. Try making sure
they're at 90 degree angles to each other and physically as distant as
possible. Try putting some steel in between them. The ARRL Handbook is a
good primer on RF construction techniques.

If copper worked that way, the makers of mu-metal would be out a lot of
money...

 

[John Smith]

"Paul Burridge" <pb@osiris1.notthisbit.co.uk> wrote in message
news:boad0091i9ta4kevuikgeu6oricaeqn9a1@4ax.com...

On Thu, 15 Jan 2004 01:55:49 GMT, "John Smith" <jocjo-john@yooha.com
wrote:

I'd bet it's oscillating and your probe is detuning it enough to
stabilize
it. Try shorting node 2 or node 3 to ground and see if you get the same
effect.

For the sake of completeness, I've shorted nodes 2,3,4 & 5 to GND
sequentially and noted the following effect on the meter needle:

N2: Slightly worse (higher meter reading)
N3: Big reduction to about 5% of FSD
N4: Total zeroing to natural resting position
N5: Reduction to about 10% of FSD.

Shorting out N4 therefore produced the most dramatic reaction but
maybe the whole thing gets killed by doing that; I don't know...
--

Let's ignore the results of shorting of nodes 3 and 5 since that will kill
the positive supply.

I'm a little surprised by the increase when shorting node 2 to common, but
then what do I know?

Parasitic oscillations are indicated, I think, at stage 2. Be sure the axes
of L1 and L2 are perpendicular to each other. L1/C2 are approximately
resonant. Try temporarily changing C2 or L1 to several times their design
value and see what happens. Did you wind your own coils? If so, you could
try spreading the turns of L1 instead.

Bypass capacitors must have short leads. Did you see the photo of my
assembly on abse? Where possible, the components have almost no leads. Is
your circuit built on a plane of copper? Try temporarily changing R2 to a
10K. If I recall accurately, someone here said something about the gain of
the transistor being low at lower collector currents. So, by reducing gain
you may achieve stability.

Pant, pant. Keep trying.

John

 

[Fred Bartoli]

"Walter Harley" <walterh@cafewalterNOSPAM.com> a écrit dans le message news:
100ek8ssb9ca55b@corp.supernews.com...

"Paul Burridge" <pb@osiris1.notthisbit.co.uk> wrote in message
newsq8d00d2fbmgv96kb8nmpmct473griocd6@4ax.com...
How are you keeping L1 from talking to L2?

I've tried interposing a grounded copper plate between them but it
made no difference. Pity as it was a good suggestion.

No, it was not a good suggestion.

Yes at 40MHz it is.

A grounded copper plate would shield against electrical field, not
magnetic
field. Proximate inductors couple via magnetic fields. Try making sure
they're at 90 degree angles to each other and physically as distant as
possible. Try putting some steel in between them. The ARRL Handbook is a
good primer on RF construction techniques.

If copper worked that way, the makers of mu-metal would be out a lot of
money...

Mumetal works better than copper for LF.

Fred.

 

[mike]

Paul Burridge wrote:

On Wed, 14 Jan 2004 19:09:42 -0800, mike <spamme0@juno.com> wrote:

snip

This isn't my design, remember. It was posted here by someone who
knows what they're talking about, was widely applauded by others who
presumably know the same, and built in a modified form by John Smith
(only one RF stage instead of two) and reported to be working fine.

Like you said, they built a DIFFERENT circuit.


"Multiple oscillations just waiting to happen" Seems you're right! How
could you tell that from just inspection of the schematic?

That's why they pay me the big bux.

When you design a circuit, it needs to be consistent with the
application and the environment.
A circuit that might be optimum in a controlled
environment when built by an experienced expert in the field may be
totally unworkable for a novice builder.

When you design a circuit, you should try to make it as independent
as possible of active device parameters and strays. This goes double
when your builder is inexperienced and doesn't know the effect that
circuit layout might have.

For a novice:
Never put a capacitor on an emitter follower.
Never do impedance transformations with Ls and Cs.
Use transformers/tapped inductors if you must, but it's
better to use circuit topologies that eliminate that need.

Yes, this stuff is done all the time...by experienced designers.

So, understand the input and output requirements.

Map out how to get the required gain and impedance transformations
with stable, reliable stages. Use resonators for selectivity and
transistors for gain. Remember that a circuit with fewer parts
is no bargain if you can't actually build it and make it work.

Design the circuit.
Only then, think about simulation. Make sure you understand how you're
going to assemble the hardware and include the influences of those stray
elements. (And that's the rub. The problems will be in the things you
didn't think to simulate. How much coupling between L1 and L2 did you
include in your simulation? Coupling thru the ground and supply leads?)
If you don't like the result, do the redesign. Don't expect the
simulation to change parameters to get you a result you didn't design
for. If you did it right, the circuit operation will depend on only
a few passive components that you have a good handle on. If your
simulation shows something you didn't expect, you should be able to
say, "Ah crap, I forgot about that...here's what I'll do to fix it."
If you just start tweeking simulation parameters, you're lost.

mike

--
Return address is VALID.
Bunch of stuff For Sale and Wanted at the link below.
Toshiba & Compaq LiIon Batteries, Test Equipment
Honda CB-125S $800 in PDX
TEK Sampling Sweep Plugin and RM564
Tek 2465 $800, ham radio, 30pS pulser
Tektronix Concept Books, spot welding head...
http://www.geocities.com/SiliconValley/Monitor/4710/

 

[John Woodgate]

I read in sci.electronics.design that mike <spamme0@juno.com> wrote (in
<4007D122.5060409@juno.com&gt about 'That Field-Strengh Meter again...',
on Fri, 16 Jan 2004:

For a novice:
Never put a capacitor on an emitter follower.

That's *exceedingly* helpful for a novice. WHERE should you not put it?

I hope you don't teach!
--
Regards, John Woodgate, OOO - Own Opinions Only. http://www.jmwa.demon.co.uk
Interested in professional sound reinforcement and distribution? Then go to
http://www.isce.org.uk
PLEASE do NOT copy news posts to me by E-MAIL!

 

[Paul Burridge]

On Fri, 16 Jan 2004 00:27:41 +0100, "Fred Bartoli"
<fred._canxxxel_this_bartoli@RemoveThatAlso_free.fr_AndThisToo> wrote:

Unfortunately that's the kind of pb that may be easily diagnosed and solved
by the knowledgeable peoples when having the beast in hands but *much*
harder at distance.

Check your proto against elementary and good wiring practice :
Ground plane, good supply decoupling/bypass, short tracks and leads
(specially for bypass caps)
low inter stage to stage parasitics coupling (capacitive, inductive and via
power rail). If you've used a ground plane, gnd ref should not be a pb for
such a simple design unless you've done a VERY VERY BIG MISTAKE.

I'm going to post a scan of the bottom (trace side) of the pcb
together with a photo of it from the component side later today. That
should give everyone a pretty good idea as to whether I've made a
balls-up of the layout.

Hmm, wait a minute. Looking at the schematics and, ahem, knowing you, how
did you handled the supply rail ?

Funny you should mention that. At places it oscillates beautifully at
15.5Mhz at around 300mV p-p!

Did you only provided a pair of 10n caps right at the batteries as shown on
the schematics ?

Yes. That's all the schematic specified.

If the answer is yes, never forget that real wires are never near from the
perfect ones in a sim.

You're right. I forgot about that.

Make sure you've carefully bypassed the nodal point between L1 and L2.
Put a 22n to 100n cap right at top of L1 and L2 with possibly a 10R resistor
between each and keep lead short.

from meter
stage supply
|
|
+---||-+----10R-----+----||-+
GND ) ) GND
)L1 )L2
) )

Thanks, Fred. Sounds like good advice. I'll post the photos first,
though, so you can see it as it stands at the moment.
--

My opinion is worth what you've paid for it.

 

[Paul Burridge]

On Fri, 16 Jan 2004 04:19:23 GMT, "John Smith" <jocjo-john@yooha.com>
wrote:

Let's ignore the results of shorting of nodes 3 and 5 since that will kill
the positive supply.

I'm a little surprised by the increase when shorting node 2 to common, but
then what do I know?

Parasitic oscillations are indicated, I think, at stage 2. Be sure the axes
of L1 and L2 are perpendicular to each other. L1/C2 are approximately
resonant. Try temporarily changing C2 or L1 to several times their design
value and see what happens. Did you wind your own coils? If so, you could
try spreading the turns of L1 instead.

Bypass capacitors must have short leads. Did you see the photo of my
assembly on abse? Where possible, the components have almost no leads. Is
your circuit built on a plane of copper? Try temporarily changing R2 to a
10K. If I recall accurately, someone here said something about the gain of
the transistor being low at lower collector currents. So, by reducing gain
you may achieve stability.

Pant, pant. Keep trying.

Thanks, John. I'm going to posts pics of the thing from above and
below so folks here can see what they're looking at. It's silly
expecting people to come up with answers when they can't see the
layout of the actual circuit. ISTR, however, that changing the 10pf
cap's value drasitcally changes the resonant frequency of the thing,
so whilst it might cure the paracitics, it ain't going to be any use
for carrying out tests on 40Mhz any more! Keep your powder dry until I
post the pix.
--

My opinion is worth what you've paid for it.

 

[mike]

John Woodgate wrote:

I read in sci.electronics.design that mike <spamme0@juno.com> wrote (in
4007D122.5060409@juno.com&gt about 'That Field-Strengh Meter again...',
on Fri, 16 Jan 2004:

For a novice:
Never put a capacitor on an emitter follower.


That's *exceedingly* helpful for a novice. WHERE should you not put it?

I hope you don't teach!

My mistake. Never put a capacitor from the emitter of an emitter
follower to ground. Better?
mike

--
Return address is VALID.
Bunch of stuff For Sale and Wanted at the link below.
Toshiba & Compaq LiIon Batteries, Test Equipment
Honda CB-125S $800 in PDX
TEK Sampling Sweep Plugin and RM564
Tek 2465 $800, ham radio, 30pS pulser
Tektronix Concept Books, spot welding head...
http://www.geocities.com/SiliconValley/Monitor/4710/

 

[James Meyer]

On Thu, 15 Jan 2004 18:45:22 -0800, "Walter Harley"
<walterh@cafewalterNOSPAM.com> posted this:

"Paul Burridge" <pb@osiris1.notthisbit.co.uk> wrote in message
newsq8d00d2fbmgv96kb8nmpmct473griocd6@4ax.com...
How are you keeping L1 from talking to L2?

I've tried interposing a grounded copper plate between them but it
made no difference. Pity as it was a good suggestion.

No, it was not a good suggestion.

A grounded copper plate would shield against electrical field, not magnetic
field. Proximate inductors couple via magnetic fields. Try making sure
they're at 90 degree angles to each other and physically as distant as
possible. Try putting some steel in between them. The ARRL Handbook is a
good primer on RF construction techniques.

If copper worked that way, the makers of mu-metal would be out a lot of
money...

This has come up twice in as many days here on SED. A copper shield
*IS* effective for medium to high frequency AC magnetic fields. Mu-metal is the
cure only for DC to low frequency magnetic fields and not any more effective for
higher frequencies than copper is.

An inductor with a very high 40 MHz current flowing through it placed in
a copper box will have a magnetic field external to the box that is so low that
nobody, not even Walter Harley, can detect it.

Jim

 

[James Meyer]

On Fri, 16 Jan 2004 04:30:21 -0800, mike <spamme0@juno.com> posted this:

John Woodgate wrote:
I read in sci.electronics.design that mike <spamme0@juno.com> wrote (in
4007D122.5060409@juno.com&gt about 'That Field-Strengh Meter again...',
on Fri, 16 Jan 2004:

For a novice:
Never put a capacitor on an emitter follower.


That's *exceedingly* helpful for a novice. WHERE should you not put it?

I hope you don't teach!

My mistake. Never put a capacitor from the emitter of an emitter
follower to ground. Better?
mike

No. That's not better. A resistor has a parasitic capacitance
associated with it. By your "rule", one should never put a resistor from the
emitter to ground either. And how about a wire from the emitter to the next
stage? A wire has a capacitance to ground in the real world. And how about the
input capacitance of the following stage its self?

My rules for a novice:

1. Never design anything.
2. Never build anything.

Jim

 

[Winfield Hill]

James Meyer wrote...

An inductor with a very high 40 MHz current flowing through it placed
in a copper box will have a magnetic field external to the box that is
so low that nobody, not even Walter Harley, can detect it.

How about aluminum with high 300kHz currents inside it?

Thanks,
- Win

whill_at_picovolt-dot-com

 

[Roger Johansson]

mike <spamme0@juno.com> wrote:

Never put a capacitor from the emitter of an emitter
follower to ground.

Because when the voltage on the base changes quickly the emitter
cannot follow quickly enough, which will cause big voltage differences
between base and emitter.

A destroyed transistor will be the result if the voltage differences
are big enough. Distortion and other problems if the differences are
smaller, depending on the circuit and the intended function.



--
Roger J.

(My email address is a spam trap, don't use it)

 

[mike]

James Meyer wrote:

On Fri, 16 Jan 2004 04:30:21 -0800, mike <spamme0@juno.com> posted this:


John Woodgate wrote:

I read in sci.electronics.design that mike <spamme0@juno.com> wrote (in
4007D122.5060409@juno.com&gt about 'That Field-Strengh Meter again...',
on Fri, 16 Jan 2004:


For a novice:
Never put a capacitor on an emitter follower.


That's *exceedingly* helpful for a novice. WHERE should you not put it?

I hope you don't teach!

My mistake. Never put a capacitor from the emitter of an emitter
follower to ground. Better?
mike


No. That's not better. A resistor has a parasitic capacitance
associated with it. By your "rule", one should never put a resistor from the
emitter to ground either. And how about a wire from the emitter to the next
stage? A wire has a capacitance to ground in the real world. And how about the
input capacitance of the following stage its self?

My rules for a novice:

1. Never design anything.
2. Never build anything.

Jim

Ok, I'll try again.
Capacitors when realized by real components in an actual
buildable circuit such as C5 and C6 in the circuit under discussion,
often cause oscillation in the presence of the inductance that CAN be
present in the equivalent circuit presented
by an emitter follower. This is exacerbated by a capacitor realized by
real components in a buildable circuit such as C4 in the circuit under
discussion.

And yes, small value parasitic capacitances present in an actual
buildable circuit as the one under discussion CAN and DO result in
oscillations.

Any time you deliberately hang a cap (to ground) on the emitter of an
emitter follower, you risk creating an oscillator.


So, come back from the deep end. The more you know about such things,
the more you instinctively create circuits without these kinds of
problems. It's stuff you don't learn in SPICE class.
mike

--
Return address is VALID.
Bunch of stuff For Sale and Wanted at the link below.
Toshiba & Compaq LiIon Batteries, Test Equipment
Honda CB-125S $800 in PDX
TEK Sampling Sweep Plugin and RM564
Tek 2465 $800, ham radio, 30pS pulser
Tektronix Concept Books, spot welding head...
http://www.geocities.com/SiliconValley/Monitor/4710/

 

[John Woodgate]

I read in sci.electronics.design that mike <spamme0@juno.com> wrote (in
<4007D95D.8020607@juno.com&gt about 'That Field-Strengh Meter again...',
on Fri, 16 Jan 2004:

My mistake. Never put a capacitor from the emitter of an emitter follower to
ground. Better?

Much better. Now, for extra credit, why not? An emitter follower looks
just right for feeding a long cable, which is a capacitive load. (;-)
--
Regards, John Woodgate, OOO - Own Opinions Only. http://www.jmwa.demon.co.uk
Interested in professional sound reinforcement and distribution? Then go to
http://www.isce.org.uk
PLEASE do NOT copy news posts to me by E-MAIL!

 

[Fred Bartoli]

"Roger Johansson" <no-email@home.se> a écrit dans le message news:
f7qf009ciumqcq4hfr02fkrna9fgusnul3@4ax.com...

mike <spamme0@juno.com> wrote:

Never put a capacitor from the emitter of an emitter
follower to ground.

Because when the voltage on the base changes quickly the emitter
cannot follow quickly enough, which will cause big voltage differences
between base and emitter.

A destroyed transistor will be the result if the voltage differences
are big enough. Distortion and other problems if the differences are
smaller, depending on the circuit and the intended function.

Nope. You'll hardly destroy any transistor with a 1n cap.
It's just because such a stage exhibits an input impedance with a negative
real part on some frequency band.
A good receipe for disaster.
If your lucky with the output impedance of the preceding stage and maybe
parasitics then you have just build an amplifier that doesn't.

This can be and is used to make oscillators.

Fred.

--
Roger J.

(My email address is a spam trap, don't use it)

 

[John Smith]

Hmm, wait a minute. Looking at the schematics and, ahem, knowing you, how
did you handled the supply rail ?

Funny you should mention that. At places it oscillates beautifully at
15.5Mhz at around 300mV p-p!

What are you doing with that 720 Ohm in your power supply? That's much too
high for your circuit. 300mV across 750 Ohms is only 0.4mA p-p current. The
source impedance is probably contributing or even causing your oscillations.
*Decouple*

You're losing about 2.8V through the 720 Ohm resistor just due to the
quiescent current of the first two stages.

If you can't replace the zeners and power supply with batteries, then at
least reduce the resistor drastically. Add a couple of 10uF tantalums around
the zeners on the board. Leave the .01nF caps on there. Add a couple of
0.1nF caps around the tantalums.

It may be a loosing cause if you don't provide a low source impedance.

John

 

[Paul Burridge]

On Fri, 16 Jan 2004 15:20:28 GMT, "John Smith" <jocjo-john@yooha.com>
wrote:

What are you doing with that 720 Ohm in your power supply? That's much too
high for your circuit. 300mV across 750 Ohms is only 0.4mA p-p current. The
source impedance is probably contributing or even causing your oscillations.
*Decouple*

You're losing about 2.8V through the 720 Ohm resistor just due to the
quiescent current of the first two stages.

If you can't replace the zeners and power supply with batteries, then at
least reduce the resistor drastically. Add a couple of 10uF tantalums around
the zeners on the board. Leave the .01nF caps on there. Add a couple of
0.1nF caps around the tantalums.

It may be a loosing cause if you don't provide a low source impedance.

Thanks, John. I've tried several times to upload the pictures but
there seems to be a problem with uploads and my ISP is looking into
it....
--

My opinion is worth what you've paid for it.

 

[mike]

John Woodgate wrote:

I read in sci.electronics.design that mike <spamme0@juno.com> wrote (in
4007D95D.8020607@juno.com&gt about 'That Field-Strengh Meter again...',
on Fri, 16 Jan 2004:


My mistake. Never put a capacitor from the emitter of an emitter follower to
ground. Better?


Much better. Now, for extra credit, why not? An emitter follower looks
just right for feeding a long cable, which is a capacitive load. (;-)

Well, assuming the resistive component of the base drive impedance
dominates, what I'd do is put an R in parallel with a C in series with the
output emitter. Match the time constant to the L/R time constant of the
output impedance. Then depending on that number, I'd either series
or shunt another resistor to match the characteristic impedance of the
cable. If you can't stand the series resistance on the output, you can
do a similar thing with a series R-C on the base, but then you need
higher drive impedance which slows things down. Emitter followers suck
at high frequencies. There ain't no free lunch.

The reason I'd do this is that a very long (unterminated) cable is NOT a
cpacitive load.
It can be APPROXIMATED to first order by a capacitor as long as the
cable is SHORT in relation to the highest frequency components to be
transmitted. A terminated cable can be approximated by a RESISTOR.

Your move.
mike

--
Return address is VALID.
Bunch of stuff For Sale and Wanted at the link below.
Toshiba & Compaq LiIon Batteries, Test Equipment
Honda CB-125S $800 in PDX
TEK Sampling Sweep Plugin and RM564
Tek 2465 $800, ham radio, 30pS pulser
Tektronix Concept Books, spot welding head...
http://www.geocities.com/SiliconValley/Monitor/4710/

 

[Jim Meyer]

Winfield Hill <Winfield_member@newsguy.com> wrote in message news:<bu8o13069b@drn.newsguy.com>...

James Meyer wrote...

An inductor with a very high 40 MHz current flowing through it placed
in a copper box will have a magnetic field external to the box that is
so low that nobody, not even Walter Harley, can detect it.

How about aluminum with high 300kHz currents inside it?

The effectiveness of shielding of AC magnetic fields by
conductors is proportional the conductivity of the shielding material.
A superconductor with zero ohms of resistance is 100% effective.
Silver is almost as good, copper next in line, and aluminum somewhat
less effective. If the field is really intense and the aluminum
really thin, then some field is going to make it through.

How hot did your aluminum shield get?

Jim